Using the Strain Index and TLV for HAL to Predict Incidence of Aggregate Distal Upper Extremity Disorders in a Prospective CohortTheses and Dissertations
August 2012
Using the Strain Index and TLV for HAL to Predict Incidence of Aggregate Distal Upper Extremity Disorders in a Prospective Cohort Tiffany Amber Cash University of Wisconsin-Milwaukee
Follow this and additional works at: https://dc.uwm.edu/etd Part of the Occupational Therapy Commons
This Thesis is brought to you for free and open access by UWM Digital Commons. It has been accepted for inclusion in Theses and Dissertations by an authorized administrator of UWM Digital Commons. For more information, please contact open-access@uwm.edu.
Recommended Citation Cash, Tiffany Amber, "Using the Strain Index and TLV for HAL to Predict Incidence of Aggregate Distal Upper Extremity Disorders in a Prospective Cohort" (2012). Theses and Dissertations. 17. https://dc.uwm.edu/etd/17
AGGREGATE DISTAL UPPER EXTREMITY DISORDERS IN A PROSPECTIVE
COHORT
by
Master of Science
in Occupational Therapy
ii
ABSTRACT USING THE STRAIN INDEX AND TLV FOR HAL TO PREDICT INCIDENT OF AGGREGATE DISTAL UPPER EXTREMITY DISORDERS IN A PROSPECTIVE
COHORT
by
The University of Wisconsin-Milwaukee, 2012 Under the Supervision of Jay Kapellusch, Ph.D.
Work-related distal upper extremity (DUE) musculoskeletal disorders (MSDs) are
very prevalent and costly in the United States. It is important to recognize working
conditions that lead to these disorders, in order to lessen the impact that they have on
workers and their employers. Identifying jobs that are likely to cause DUE MSDs is
difficult because there are many factors that are believed to contribute to DUE MSD
development. The current study aims to determine if the Strain Index (SI) and the
ACGIH TLV for HAL (two DUE job physical exposure assessment methods) predict
increased risk of workers developing aggregate DUE MSDs. For this study, aggregate
disorders include: (i) carpal tunnel syndrome, (ii) lateral epicondylitis, (iii) medial
epicondylitis, (iv) tendonitis of wrist flexors and extensors, (v) de Quervain’s disease,
and (vi) trigger finger.
Subjects for this study were drawn from a recently completed large-scale
prospective cohort study consisting of 1,205 volunteer workers from 21manufacturing
companies located in IL, UT, and WI. Of the 1,205 workers, only those workers who had
no previous history of an aggregate disorder at study onset will be considered. Workers
were followed monthly to determine if new DUE MSD symptoms developed. Specific
case definitions are used to identify when a worker develops one or more aggregate DUE
iii
MSD. Physical exposures from workers’ jobs were individually measured and videos
were recorded at baseline. Jobs were investigated quarterly to determine physical
exposure changes and re-analyzed as necessary. Time to first aggregate DUE MSD was
modeled using proportional hazards regression to determine if there is a relationship
between SI and TLV for HAL scores and increased risk of developing DUE MSDs while
controlling for relevant covariates (age, gender, BMI).
Univariate analyses, showed a strong relationship between age (HR = 1.03, p =
0.001) and gender (HR = 2.38, p = 0.002) and the development of aggregate DUE MSDs.
There was suggestive evidence that the SI, with a cut point of 6.1 (p = 0.13), predicts
increased risk of first lifetime aggregate DUE MSD. No significance was noted for the
TLV for HAL. Efforts per minute showed a slightly significant association using a spline
placed at 37.3 (p = 0.03). Multivariate analyses found suggestive evidence for an
association between efforts per minute when analyzing using a spline placed at 37.3
efforts per minute (p = 0.08). No effect was found with the SI or TLV for HAL.
Age and gender appear to be significantly associated with the development of first
lifetime DUE MSD. The SI appears to be a more reliable method to use to determine
jobs that place workers at increased risk of developing first lifetime aggregate DUE
MSD, when comparing it to the TLV for HAL.
iv
Carpal Tunnel Syndrome………………………………………………….4
De Quervain’s Disease…………………………………………………...13
Trigger Finger……………………………………………………………15
Brief Description of Parent Study………………………………………..26
Baseline Health Data – Description……………………………………...26
Baseline Job Data – Description…………………………………………27
Follow-up Assessment of UED Health – Description……………...……27
Follow-up Job Data………………………………………………………28
Methodology of Current Research……………………………………………….28
Determination of Study Cohort…………………………………………..28
Computation of ‘Job Metrics’ at the Task and Job Levels………………29
Determining Scores for the Strain Index and TLV for HAL…………….30
Occupational UED Health Outcomes – Case Definitions.………………31
v
Results……………………………………………………………………………………36
Covariates...………………………..…………………………………….38
Associations between Physical Exposure and First Aggregate DUE MSD……..49
Age, Gender, and BMI as Risk Factors for First Aggregate DUE MSD………...52
Differences among the Virgin and Prevalent Cohorts…………………………...53
Development of MSDs in a Virgin Cohort………………………………………54
Strengths and Weaknesses of the Study…………………………………………54
Future Studies……………………………………………………………………56
vi
Appendix B: Force Algorithm…………………………………………………..78
vii
LIST OF FIGURES
Figure 1. Example of how the breakdown of job, tasks, and subtasks occurred for each
worker………………………………………………………………………………..…29
LIST OF TABLES
Table 1: Table 1: Specific case definitions for aggregate disorders……………………32
Table 2: Point and lifetime prevalence for the six aggregate DUE MSDs………...……37
Table 3: Order of occurrence for specific aggregate disorders……………………...…..38
Table 4: Descriptive statistics for covariates…………………………………..………..39
Table 5: Descriptive statistics for physical exposure factors……………………………40
Table 6: Univariate hazard ratios for covariates………………………………………...42
Table 7: Univariate hazard ratios for exposure variables……………………………….42
Table 8: Multivariate model for risk of aggregate disorders with analyst SI variable.....44
Table 9: Multivariate model for risk of aggregate disorders with analyst SI variable with
2 categories……………………………………………………………………….……...45
Table 10: Multivariate model for risk of aggregate disorders with algorithm SI
variable………………………………………………………………………….………..45
Table 11: Multivariate model for risk of aggregate disorders with algorithm SI variable
with 2 categories…………………………………………….……………………...……46
Table 12: Multivariate model for risk of aggregate disorders with analyst TLV for HAL
variable………………………………………………………………………………..…46
Table 13: Multivariate model for risk of aggregate disorders with analyst TLV for HAL
with 3 categories…………………………………………………………………….…...47
ix
Table 14: Multivariate model for risk of aggregate disorders with efforts per minute
variable…………………………………………………………………………………...47
Table 15: Multivariate model for risk of aggregate disorders with efforts per minute
variable with spline………………………………………………………………………48
1
Using the Strain Index and TLV for HAL to Predict Incidence of Aggregate Distal Upper
Extremity Disorders in a Prospective Cohort
Despite great attention from researchers and practitioners, work-related
musculoskeletal disorders (WMSDs) remain common and costly to industry. In 1996, the
National Institute for Occupational Safety and Health (NIOSH) estimated the cost to be
$13 billion annually, in 1997 the American Federation of Labor and Congress of
Industrial Organizations (AFL-CIO) estimated it to be $20 billion annually, and in 2001
the National Research Council and the Institute of Medicine estimated it to be from $45
to $54 billion annually. In 2008, it was estimated that WMSDs cost U.S. industry $53.42
billion in direct U.S. workers compensation costs, annually. (Liberty Mutual Workplace
Safety Index, 2010) According to the most recent Bureau of Labor Statistics report
(2010) there were 284,340 work-related musculoskeletal disorders (WMSDs) in the
United States. Of these, 78.2% resulted in lost time. The most often injured body region
is the back, contributing 47% of all injuries, followed by distal upper extremity with
14.4% and shoulder with 13.3% (BLS, 2010). The occupations with the highest
incidence of WMSDs occur in service occupations (70,780), followed by transportation
and material moving occupations (58,060), and then production occupations (33,280).
The incident rate for WMSDs was 34 cases per 10,000 full-time workers in 2010 and this
number has remained virtually unchanged from the last 20 years. (Bureau of Labor
Statistics, 2010, 2008, 2000)
WMSDs are also costly to workers, commonly creating a loss in income for
injured workers, mostly due to days missed from work. (Spreeuwers, de Boer Verbeck,
van Beurden, de Wilde, Braam, Willemse, & van Dijk, 2011) According to the Bureau of
2
Labor Statistics (2010), lost time WMSDs averaged 10 days away from work annually.
For the distal upper extremity (DUE), hand/wrist injury in general resulted in 14 lost
days, with more prevalent disorders, such as carpal tunnel syndrome (CTS) and tendonitis
resulting in 27 and 15 lost days on average, respectively.
Employers with frequent and/or severely injured workers often suffer losses in
productivity and worker skill. For example, Martimo, Shiri, Miranda, Ketola, Varonen,
& Viikari-Juntura (2009), found that 56% of workers with an upper extremity disorder
reported a productivity loss, with an average reduction in productivity of about 34%. A
study by Keogh, Nuwayhid, Gordon, & Gucer (2000) found that 53% of patients who had
acquired a work-related DUE MSD and who had claimed compensation, reported
persistent symptoms that were severe enough to interfere with work during four years
post-claim, leading to a loss in productivity for the employer.
WMSDs have a complex, multi-factorial etiology (Bernard, 1997), including
individual risk factors (e.g. age, gender, BMI, etc.), psychosocial factors (e.g. job
satisfaction, job control, etc.), and job physical exposure (e.g. force, frequency, etc.).
Various job analysis methods have been developed that attempt to address the multi-
factorial etiology of WMSDs. However, none have been completely validated and all
have limitations.
This study quantifies job physical exposures using the SI and TLV for HAL to
determine if physical exposure is associated with increased risk of aggregate DUE MSDs
while controlling for age, gender, and obesity (measured using body mass index (BMI)).
3
The study’s specific hypothesis is that there is a relationship between scores of (i) the
Strain Index (SI) and (ii) the American Conference of Governmental Industrial
Hygienists Threshold Limit Value for Hand Activity Limit (TLV for HAL) and incidence
of aggregate upper extremity disorders.
4
This study considers CTS, lateral epicondylitis, medial epicondylitis, hand/wrist
flexor and extensor tendonitis, de Quervain’s disease, and trigger finger as aggregate
DUE MSDs. The Occupational Safety and Health Administration (OSHA) defines
MSDs as “disorders of the muscles, nerves, tendons, ligaments, joints, cartilage and
spinal discs.” They specify that MSDs “do not include disorders caused by slips, trips,
falls, motor vehicle accidents, or other similar accidents.” The following summaries are
of the etiology and diagnosis criteria of the specific aggregate disorders, followed by
landmark studies that describe specific risk factors for each disorder.
Carpal Tunnel Syndrome. CTS is an entrapment neuropathy that is caused by
the compression of the median nerve as it passes through the carpal tunnel of the wrist.
This compression is caused by increased intra-tunnel pressure. (Herbert, Gerr, &
Dropkin, 2000) Signs and symptoms include: (i) numbness/tingling in two or more
median nerve served digits (1-4) and (ii) an abnormal nerve conduction study (NCS)
(Harrington, Carter, Birrell, & Gompertz, 1998).
A study by Silverstein, Fine, & Armstrong (1987) analyzed risk factors for CTS
with 652 workers from 7 different industrial sites. The workers were classified into four
groups based on their level of physical exposure: high force-high repetitiveness, high
force-low repetitiveness, high repetitiveness-low force, and low force-low repetitiveness.
The authors found that the prevalence for CTS ranged from 0.6% among workers in low
force-low repetitive jobs to 5.6% among workers in high force-high repetitive jobs. With
gender, age, and years on the job analyzed as confounders, the authors found that CTS
5
was strongly associated with high force-high repetitive jobs (OR = 15.5, p < 0.001) when
compared to low force-low repetitive jobs. In addition the authors found, to a lesser
extent, low force-high repetitive jobs (OR = 2.7) and high force-low repetitive jobs (OR
= 1.8) to be associated with CTS, but not statistically significant. The authors found that
repetitiveness appeared to be a stronger risk factor than force (OR = 5.5, p < 0.05).
Gender, age, and years on the job were not statistically significant. Therefore, jobs that
include high efforts per minute or high efforts per minute and a high intensity of exertion
are risk factors for CTS.
Though considered a landmark study for determining risk factors associated with
CTS, Silverstein, et. al. (1987) has a couple weaknesses. One limitation is that the study
was retrospective, relying on subject recall of date of onset to determine if the CTS had
originated while participating in the study. Subject recall is not a highly reliable method
of determining CTS prevalence. Another limitation is that survivor/selection bias
occurred due to only accepting active workers with at least one year on the job, in order
to exclude workers with less seniority because they may have brought CTS previously
obtained on another job to the one under study. This may have excluded the workers
with less seniority, who had acquired CTS during the study, which would decrease the
study’s CTS prevalence. Another limitation is that NCSs were not used for the diagnosis
of CTS, which could have led to an overestimation of CTS prevalence.
A study by Violante, Armstrong, Fiorentini, Graziosi, Risi, Venturi, Curti,
Zanardi, Cooke, Bonfiglioli, & Mattioli (2007) assessed risks associated with work-
related biomechanical overloads in the onset and course of CTS. To accomplish this, the
authors evaluated 2,092 workers in work-groups with job tasks spanning different
6
biomechanical exposures at baseline in terms of ACGIH hand-activity/peak force action
limit and TLV. The authors found that one-year incidence of CTS symptoms was 7.3%.
“Unacceptable” overload was associated with a 3-fold increased risk of onset of CTS
symptoms as compared to “acceptable” load. Workers who experienced “borderline
overload” appeared to be associated with a 1.5-fold increase in risk. Female gender was
a stronger risk factor among exposed workers (OR = 7.3, in workers with “unacceptable
overload”, OR = 6.7, in “borderline overload” vs OR = 2.5 in “acceptable load”). Being
overweight/obese appeared to be an independent risk factor among exposed workers (OR
= 1.9 in workers with “unacceptable” overload, OR = 1.7 to 2.8 for “borderline” overload
vs OR = 1.1 to 1.5 in “acceptable load”). Therefore, the authors came to the conclusion
that risk factors for CTS are jobs that require a biomechanical overload. They also found
the female gender and increased BMI to be risk factors.
The Violante, et al. (2007) study has multiple strengths and a few limitations. A
strength of the study is that the authors used an agreed upon case definition and NCSs to
determine CTS diagnosis, in order to be confident in the diagnoses. Additionally, a large
cohort of workers was used in a broad range of occupations, which makes the results
generalizable. However, the authors did not break down the type of physical exposures
seen on the job, so the reader does not know more about the specific type of exposures
that led to the development of CTS.
Lateral Epicondylitis. Lateral epicondylitis is caused by a lesion at the common
extensor origin of the lateral epicondyle of the humerus. (Harrington, et. al., 1998) In
addition to lesions, vascular hyperplasia and active fibroblasts can also occur at the
common extensor origin of the lateral epicondyle, which contributes to the development
7
of this disorder (Van Hofwegen, Baker, & Baker, 2010). Vascular hyperplasia and active
fibroblasts are the pathologic healing response to microtears caused by repetitive
eccentric or concentric overloading of the extensor muscle mass. (Van Hofwegen, et al.,
2010) Signs and symptoms for lateral epicondylitis are: (i) lateral elbow pain, (ii) pain
upon palpation of one or more of six lateral tender points, and (iii) a positive resisted
wrist extension test, which is a maneuver of bending the wrist backward against
resistance (extension) causing pain for those with lateral epicondylitis (Harrington, et. al.,
1998).
A case-referent study by Haahr & Andersen (2003) suggests that physical
exposure factors are associated with lateral epicondylitis. The authors compared 267 new
cases of lateral epicondylitis to 388 referents and found that manual job tasks (OR = 3.1,
95% CI = 1.9 – 5.1), self-reported “posture”, (arms lifted away from body: males OR =
2.1, 95% CI = 1.1 – 4.3, females OR = 4.4, 95% CI = 2.3 – 8.3; hands bent or twisted:
males OR = 3.2, 95% CI = 1.5 – 6.9, females OR = 10.0, 95% CI = 4.1 – 22.4) and
“forceful work” (males OR = 2.2, 95% CI = 1.3 – 3.9, females OR = 2.8, 95% CI = 1.6 –
5.0) were related to lateral epicondylitis. The authors also found that among females,
work involving performing repeated movements of the arms was related to lateral
epicondylitis (OR = 3.7, 95% CI = 1.7 – 8.3). Among males, the authors found that work
with precision demanding movements was related to lateral epicondylitis (OR = 5.2, 95%
CI = 1.5 – 17.9). An index was established based on posture, repetition, and force. The
adjusted ORs for lateral epicondylitis at low, medium, and high index values were 1.4
(95% CI = 0.8 – 2.7), 2.0 (95% CI = 1.1 – 3.7), and 4.4 (95% CI = 2.3 – 8.7). The
authors conclude that physical exposure factors, such as manual job tasks, posture,
8
intensity of exertion, efforts per minute, and work requiring precision movements are risk
factors for lateral epicondylitis.
One limitation of the Haahr & Andersen (2003) study is that selection bias
occurred due to the recruitment of only cases attending general practice. These patients
may be those with more severe symptoms or those experiencing the greatest problems in
performing their activities of daily living. In addition, an information bias could have
occurred due to information on job exposure having been collected by questionnaire,
which obtains limited and subjective information.
Medial Epicondylitis. Similar to lateral epicondylitis, medial epicondylitis is
caused by a lesion at a common muscle group origin, but medial epicodylitis occurs at the
medial epicondyle of the humerus, where the flexor wrist and hand muscles originate.
(Harrington, et. al., 1998) The pathologic healing responses, mentioned with lateral
epicondylitis, of vascular hyperplasia and active fibroblasts, can also occur with this
disorder (Ciccotti, Schwartz, & Ciccotti, 2004). Signs and symptoms required for the
diagnosis of medial epicondylitis are: (i) medial elbow pain, (ii) pain upon palpation of
one or more of two medial tender points, and (iii) positive resisted wrist flexion, which is
a maneuver that includes flexing the wrist against resistance which causes pain at the
medial aspect of the elbow for individuals with the disorder (Mani & Gerr, 2000).
Descatha, Leclerc, Chastang, Roquelaure, & the Study Group on Repetitive Work
(2003) analyzed medial epicondylitis independently and its links between individual and
occupational risk factors in repetitive work, using a cross-sectional design. The authors
used 1,757 workers for the study, who were examined by an occupational health
physician during one year and 598 workers were again examined three years later. The
9
authors found that the prevalence for medial epicondylitis was between 4 and 5%, with
an annual incidence estimate at 1.5%. The authors determined that forceful work was a
risk factor (OR = 1.95, p = 0.01), but not exposure to repetitive work. The authors found
that workers diagnosed with medial epicondylitis had a significantly higher prevalence of
other WMSDs (at least one WMSD, p < 0.001). Therefore, a high intensity of exertion
and the prevalence of other WMSDs increase risk of developing medial epicondylitis.
The Descatha, et. al. (2003) study used a cross-sectional design, which is a
weaker design because the reader is not able to infer the temporal sequence between the
exposure and the disorder. These studies also are of a weaker design because they only
include current and not former workers; therefore, the results may be influenced by the
selective departure of workers who had already acquired a DUE MSD. Due to the
study’s design, a possible selection bias could have occurred in two ways: (i) more
occupational physicians from the firms with higher prevalence of upper-limb disorders
participated in the follow-up component of the study (which would have increased the
prevalence rate) and (ii) 102 workers were lost to follow-up (it is uncertain what
happened with these participants).
Shiri, Viikari-Juntura, Varonen, & Heliövaara (2006), attempted to estimate the
prevalence of lateral and medial epicondylitis and to investigate their risk factors. This
study analyzed 4,783 participants. The authors found the prevalence of definite lateral
epicondylitis was 1.3% and that of medial epicondylitis was 0.4%. They found that the
prevalence did not differ between men and women and was highest in subjects aged 45-
54 years. The authors found an interaction between repetitive movements of the arms
and forceful activities for the risk of possible or definite lateral epicondylitis (for both
10
repetitive and forceful activities vs. no such activity: OR = 5.6, p = 0.002). The authors
found that women’s obesity (OR = 2.7, 95% CI = 1.2 – 6.0), repetitive movements of the
arms among women (OR = 1.7, 95% CI= 1.0 – 2.9), and forceful activities among men
(OR = 2.2, 95% CI = 1.0 – 4.7) independently of each other showed significant
associations with medial epicondylitis. Therefore, the authors found different risk factors
for each disorder. Risk factors for lateral epicondylitis include jobs that require both a
high amount of efforts per minute and a high intensity. Risk factors for medial
epicondylitis include women’s obesity, efforts per minute among women, and high
intensity exertions among men. The authors conclude that physical exposure is a risk
factor for both lateral and medial epicondylitis.
A limitation of the Shiri, et al. (2006) study is that it is a cross-sectional design.
In addition, the prevalence of epicondylitis may have been underestimated and the
estimated odds ratios may have been lessened because the subjects who were not
included in the study were more frequently exposed to forceful activities than those
included in the study.
Hand/Wrist Flexor and Extensor Tendonitis. Tendonitis is caused by forces
that exceed the ability of tendinous tissue to adapt, which causes the tendon(s) to become
inflamed. (Piligian, Herbert, Hearns, Dropkin, Lansbergis, & Cherniack, 2000)
Tendonitis of the hand/wrist extensors is caused by inflammation of an extensor tendon
and tendonitis of the flexors is caused by inflammation of a flexor tendon. Tendonitis of
the hand/wrist extensors is diagnosed by noting the following signs and symptoms: (i)
dorsal wrist pain, (ii) 2-6 extensor compartment tenderness (dorsal wrist area), and (iii)
pain worsened by resisted wrist or finger extension. Tendonitis of the hand/wrist flexors
11
is diagnosed by noting these signs and symptoms: (i) volar wrist pain, (ii) no
numbness/tingling in digits 1-4, (iii) three locations of digital flexor tendon tenderness,
and (iv) positive resisted wrist flexion, which is positive if it elicits pain. (Mani & Gerr,
2000; Pilgian, et. al., 2000)
Silverstein, Fine, & Armstrong (1986) analyzed forceful and repetitive job
attributes to determine whether they were positively associated with cumulative trauma
disorders (CTDs). The authors describe CTDs as tendon-related disorders of the hand
and wrist that produce inflammation of the tendons or compression of the peripheral
nerves. Hand/wrist tendinitis is often considered a CTD. The authors analyzed a total of
574 workers from six different industrial sites that were categorized into four force
repetitive exposure groups. Significant positive associations were observed between
hand wrist CTDs and high force-high repetitive jobs (OR = 30.3, p < 0.0001), which were
independent of age, gender, years on the job, and plant. When force (low, high),
independent of repetitiveness, was entered into the model as the only exposure measure,
the odds ratio for high force was 4.4 (p < 0.0001). When repetitiveness (low, high),
irrespective of force, was entered into the model as the only exposure variable, the odds
ratio was 2.8 (p < 0.005). This study demonstrates that exertion intensity and efforts per
minute, together and independently, are risk factors for developing flexor or extensor
hand/wrist tendinitis.
A limitation of the Silverstein, et. al. (1986) article is that the results may have
underestimated the prevalence of hand wrist CTDs, due to subject selection being limited
to active workers. Additionally, the one year seniority criteria for subject selection
12
excluded those who might have had CTDs and transferred before one year, as well as
those with CTDs but were not on the job for at least one year.
A longitudinal study by Leclerc, Landre, Chastang, Niedhammer, Roquelaure, &
the Study group on Repetitive Work (2001) analyzed individual and physical exposure
risk factors associated with the development of wrist tendonitis. The authors used 598
workers in five activity sectors. The participants were given questionnaires and a
physical exam, once in two consecutive years, and then again three years later. The
authors found that the presence of somatic problems (3.78, p ≥ 0.15) and social support at
work (OR = 2.49, p ≥ 0.15) was a strong predictor of wrist tendinitis. They also found
that a BMI increase of ≥2 kg/m² (OR = 2.2, p ≥ 0.15) was associated with the incidence
of wrist tendinitis. Additionally, the authors found that workers, who reported that they
had to repetitively hit during work (OR = 2.16, p ≥ 0.15), were associated with a higher
incidence of wrist tendonitis. Therefore, somatic problems, lack of social support,
increased BMI, and repetitively hitting an object during work were risk factors for wrist
flexor or extensor tendinitis.
Although Leclerc, et. al. (2001) study points out important possible risk factors
associated with the development of wrist tendonitis, it has some limitations. The authors
note that they had difficulty interpreting results about the incidence of specific disorders
in a group in which many workers are already affected at the beginning. This is due to
the fact that the “healthy” workers at baseline represented a select group, since they were
unaffected despite a high level of occupational exposure. Another limitation is that the
temporal aspects of causality are unknown. The authors did not know the time lag
between occupational exposure and the onset of the upper-limb disorder. Therefore, it is
13
more difficulty to develop a cause and effect relationship between the exposure and the
disorder.
De Quervain’s Disease. De Quervain’s disease is caused by thickening of the
fibrous sheath or retinaculum in the first dorsal extensor compartment. (Barton, Hooper,
Noble, & Steel, 1992) The first extensor compartment is located where the extensor
pollicis brevis and the adductor pollicis longus are housed. Signs and symptoms required
to diagnose de Quervain’s disease are: (i) radial wrist pain centered over the radial
styloid, (ii) first extensor compartment tenderness (base of thumb), and (iii) a positive
Finkelstein test, which is a maneuver that includes positioning the person’s thumb within
their flexed fingers and then the hand is manipulated into ulnar deviation (test is positive
if radial wrist pain or tenderness is elicited). (Harrington, et. al., 1998; Mani & Gerr,
2000; Witt, Pess, & Gelberman, 1991)
Moore (1997) reviews information from the medical literature on occupational
risk factors associated with the development of De Quervain’s tenosynovitis. The author
noted that De Quervain’s disease tends to appear more in females than in males and
primarily between the ages of 35 and 55. The review notes that cases tend to appear
more in individuals who use their thumbs a great deal. Also, cases tend to appear more
with workers who complete fast, repetitive manipulations, or where the posture of the
hand was such that unremitting, or repetitive pinching, grasping, pulling, or pushing was
necessary. Therefore, gender manual work that involves the thumb, efforts per minute,
and posture may be risk factors for de Quervain’s disease.
Moore (1997) also notes limitations that appear in the literature regarding de
Quervain’s disease. One limitation is that few epidemiological studies exist that focus
14
specifically on de Quervain’s tenosynovitis. Moore (1997) goes on to note that there are
no studies that establish or fail to confirm an association between hand usage, including
hand usage at work, and de Quervain’s as a specific disorder. Therefore, more research
needs to be conducted to determine true risk factors for de Quervain’s disease.
A cross-sectional study by LeManac’h, Roquelaure, Ha, Bodin, Meyer, Bigot,
Veaudor, Descatha, Goldberg, & Imbernon (2011) analyzed personal and occupational
risk factors for De Quervain’s disease in a working population of 3,710 workers. Of
these participants, 45 workers were diagnosed with De Quervain’s disease by physical
examination. A standardized physical and a self-administered questionnaire were used to
assess individual factors and work exposure. The authors found that the prevalence rates
of De Quervain’s disease for the whole working population was 1.2%. The personal risk
factors that they found for De Quervain’s disease were age (OR = 1.1 for 1-year increase
with age, p = 0.001) and female gender (OR = 4.9, p = <0.001). The work-related
factors, that the authors found, were workpace dependent on technical organization (OR
= 2.0, p = 0.045), repeated or sustained wrist bending in extreme posture (OR = 2.6, p =
0.010), and repeated movements associated with the twisting or driving of screws (OR =
3.4, p = 0.001). This study demonstrates that age, gender, technical organization,
repeated or sustained wrist bending in extreme posture, and repeated twisting or driving
of screws are risk factors for de Quervains disease.
LeManach’h, et al. (2011) has a few limitations. One limitation of the
LeManach’h, et. al. (2011) study is that the authors did not exclude participants who have
been diagnosed with osteoarthritis or hand/wrist tendinitis. This may have led to an
overestimation of the amount of workers who actually had De Quervain’s disease.
15
Additionally, a healthy worker effect may have occurred, due to the study’s cross
sectional design, which may have led to an underestimation of the estimates of risk.
Trigger Finger. Trigger finger is caused by hypertrophy of the retinacular sheath
and peritendinous tissue in the volar aspect of the hand or swelling of the finger tendon,
which progressively restricts the motion of the flexor tendon. (Newport, Lane, &
Stuchin, 1990; Rozental, Zurakowski, & Blazar, 2008; Sampson, Badalamente, Hurst, &
Seidman, 1991) Thickening of the sheath, along with some localized tendon thickening,
may create a narrowed tunnel for tendon excursion and lead to a block in movement.
Trigger finger typically occurs at the site of the A1 pulley, which is located around the
area of the metacarpalphalangeal (MCP) joints (Akhtar, Bradley, Quinton, & Burke,
2005). Signs and symptoms for the diagnosis of trigger finger are: (i) pain in the finger
and focal tenderness over A-1 pulley, and (ii) demonstrated triggering (a catching of a
digital flexor tendon as it glides under the A-1 pulley. (Makkouk, Oetgen, Swigart, &
Dodds, 2008; Mani & Gerr, 2000)
There have been few studies that address the etiological factors associated with
trigger finger. Trezies, Lyons, Fielding, & Davis (1998) investigated the occupational
histories of 178 patients with diagnosed trigger finger. The authors used a questionnaire
that asked about each patient’s employment during the last 10 years and then the authors
divided their occupations were divided up into one of four categories: (i)
unemployed/housewife/retired, (ii) office work, (iii) light manual, and (iv) heavy manual.
The authors compared the histories with the 1991 Census data and found that the
distribution of their occupations was not significantly different from the local general
population, meaning that trigger finger appears to be unrelated to work.
16
The results from Trezies, et al. (1998) may not be reliable due to multiple study
limitations. A questionnaire was used to gather participant information, which is a
subjective way of acquiring information and may not be very reliable or accurate.
Additionally, the four exposure groups were classified in very generic categories, which
could encompass many different tasks. Therefore, physical exposure may have varied
greatly within each category, which would make the results less accurate.
There have been a few articles that list possible risk factors associated with the
development of trigger finger, but do not provide any direct evidence of associations. An
article by Thorson & Szabo (1989) reports that possible occupational factors that may
lead to the development of trigger finger are repetition while in in non-neutral posture,
vibration, low temperature, pressure from hard objects, forceful blows, or torques.
Another article by Rosenthal (1987) suggests that osteoarthritis, using vibrating tools,
sustained and repetitive grasps, repetitive crimping, use of small tools, or a significant
change in the customary pattern and exceptional level of hand activities may be related to
the development of trigger finger.
The above studies demonstrate three main groups of risk factors that need to be
considered when analyzing DUE MSDs: (i) physical (force, frequency, posture, etc.), (ii)
psychosocial (increased stress, limited job satisfaction, etc.), and (iii) individual (age,
gender, BMI, etc.). Most job analysis methods use physical risk factors to estimate level
of risk associated with the job. However, there is ample evidence in the literature that
suggests individual risk factors are associated with the development of DUE MSDs.
Psychosocial risk factors are often mentioned as risks, but little evidence is available
showing these factors as predictors of DUE MSDs.
17
In summary of the above studies, physical risk factors appear to have a great
amount of evidence to support their relationship with the development of DUE MSDs. A
study by Melchior, Roquelaure, Evanoff, Chastang, Ha, Imbernon, Goldberg, Leclerc,
and the Pays de la Loire Study Group (2006) analyzed the role that physical risk factors
have on the development of upper extremity MSDs among 2,656 workers. The authors
compared manual and non-manual workers. Among physical risk factors, the authors
found that repetitive movements, forceful movements, exposure to vibrations, and wrist
flexion to be significant (p < 0.0001). Additionally, a study by Moore, Rucker, & Knox
(2001) analyzed 56 jobs in order to determine the impact of various physical risk factors
on DUEs. They found repetition, gloves, and forcefulness to have an odds ratio (OR) of
≥ 9.0, p ≤ 0.01. The authors also found several significant interactions among physical
risk factors, which include interactions among high repetitiveness and high forcefulness
(OR = 27.0, p < 0.001), and high forcefulness and non-neutral posture (OR = 3.3, p =
0.03). Several studies have provided evidence that force (Descatha, et al. (2003); Haahr
& Andersen (2003); Melchior, et al. (2006); Moore, et al. (2001); Shiri, et al. (2006);
Silverstein (1987); Silverstein, et al. (1986)), repetition (Haahr & Andersen (2003);
Melchior, et al. (2006); Moore, et al. (2001); Shiri, et al. (2003); Silverstein, et al.,
(1987); Silverstein, et al. (1986)), exposure duration (Haahr & Andersen (2003); Shiri, et
al. (2006); Silverstein, et al. (1987); Silverstein, et al. (1986); Violante, et al. (2007)), and
posture (Haahr & Andersen (2003); LeManac’h, et al. (2011); Melchior, et al. (2006)) are
risk factors for DUE MSDs. However, there are a few studies that suggest physical
exposure has an uncertain relationship with DUE MSDs (Nathan, Keniston, Myers, &
Meadows (1992); Trezius, et al. (1998)).
18
According to a report by Bernard (1997) there are several individual risk factors
that need to be addressed when assessing influences on the development of WMSDs.
The three most commonly mentioned individual risk factors appear to be age, gender, and
BMI. There have been several studies that report age to be a major contributing factor to
the development of MSDs (English, Maclaren, Court-Brown, Hughes, Porter, & Wallace,
1995; Fan, Silverstein, Bao, Bonauto, Howard, Spielholz, Smith, Polissar, & Viikari-
Juntura, 2009; Ohlsson, Hansson, Balogh, Strömberg, Pålsson, Nordander, Rylander, &
Skerfving, 1994) However, Bernard (1997) mentions that a survival bias may occur when
analyzing the impact of age. Survivor bias happens when a worker develops a WMSD or
some health problem and leaves their job to take a less strenuous job, thereby leaving
only the workers who have not been negatively affected by their job (Bernard, 1997).
Multiple studies report a higher prevalence of MSDs in women than in men (Bernard,
Sauter, Fine, Petersen, & Hales, 1994; Chiang, Ko, Chen, Yu, Wu, & Chang, 1993; Fan,
et al., 2009; Hales, Sauter, Peterson, Fine, Putz-Anderson, Schleifer, Ochs, & Bernard,
1994; Johansson, 1994; Stevens, Sun, Beard, O’Fallon, & Kurland, 1988). There has also
been a lot of evidence that workers who are obese (BMI>29) tend to develop more
WMSDs when compared to those who are slender (BMI<20) (Nathan, Keniston, Myers,
& Meadows, 1992; Nathan, Keniston, Meadows, Lockwood, 1994; Nordstrom, Vierkant,
DeStefano, & Layde, 1997; Vessey, Villard-Mackintosh, & Yeates, 1990; Werner,
Albers, Franzblau, & Armstrong, 1994).
Job Analysis Methods
Commonly used job analysis methods for the DUE include the TLV for HAL, the
SI, the Rapid Upper Limb Assessment (RULA), the State of Washington Checklist, and
19
the Ergonomic Job Measurement System (EJMS). The most commonly used models in
research studies appear to be the TLV for HAL and the SI, which is the reason they are
investigated in this study. The other methods have had limited investigation into their
effectiveness as risk prediction models and is why they are not discussed in this study.
Strain Index. The SI is an assessment method for identifying those jobs that are
unsafe and likely associated with DUE MSDs and those jobs that are not (Moore and
Garg, 1995). The SI relies on the measurement or estimation of six semi-quantitative
task variables that describe the physical stress of a job based on physiological and
biomechanical theories of the DUE and epidemiological findings (Garg & Kapellusch,
2011). The six task variables used in the SI are: (i) intensity of exertion (applied force),
(ii) number of exertions per minute, (iii) percent duration of exertion per cycle, (iv)
hand/wrist posture, (v) speed of work, and (vi) duration of exposure per day. Each of
these categories has five rating values that are used to describe the physical exposure.
Multipliers correspond to the task variable ratings, which act as penalties. The
multipliers are used to compute a multiplicative score, which is the Strain Index Score.
(Moore & Garg, 1995)
A study by Moore, et al., (2001) analyzed the performance of the SI and
compared it to several risk factors. Several generic risk factors were included in the
analysis (i.e. high forcefulness, high repetitiveness, pinch grasp, gloves, non-neutral
posture, vibration, localized compression, cold, etc.). The authors found that the odds
ratio for the SI was 108.3 (CI = 16.7, 705.0) and was 3 to 16 times larger than any other
factors studied. The SI force rating alone offered the next highest odds ratio of 36.0 (CI
= 4.3, 303.4), followed by the combination of SI force rating combined with high
20
repetitiveness with an odds ratio of 31.2 (CI = 3.7, 262.1). The authors found that the SI
performed better than any of the individual or combinations of generic risk factors and
that it’s sensitivity, specificity, positive predictive value, and negative predictive value
were all approximately 0.90. These results provide evidence of the SI’s external and
predictive validity.
Another study by Rucker and Moore, 2002 analyzed predictive validity of the SI
in two manufacturing plants. Investigators, who were blinded to health outcomes,
analyzed the right and left sides of 28 jobs using the SI and classified them as
“hazardous” or “safe”. The occurrence of DUE MSDs were determined using OSHA 200
logs. When the authors compared sides, symmetry between morbidity and hazard
classification was required. When comparing jobs, this symmetry was not required. 2 x
2 contingency tables were used to determine an association between the hazard
classifications and the morbidity classifications for the 56 sides and 28 jobs. For the
sides, the authors found a significant association between hazard classification and
morbidity classification with an empirical odds ratio of 73.2. The sensitivity, specificity,
positive predictive value, and negative predictive value were 1.00, 0.84, 0.47, and 1.00,
respectively. In addition, a similar association was found with jobs, with an empirical
odds ratio of 106.6, and the sensitivity, specificity, positive predictive value, and negative
predicative value of 1.00, 0.91, 0.75, and 1.00, respectively. These results demonstrate
that the SI is able to identify tasks that place workers at increased risk of DUE MSDs.
They also demonstrate the SI’s external validity.
Knox & Moore (2001) analyzed the predictive validity of the SI in turkey
processing. Investigators, who were blinded to health outcomes, analyzed the right and
21
left sides of workers in 28 jobs using the SI and classified them as “hazardous” or “safe,”
based on the SI score. OSHA 200 logs were used to determine the occurrence of DUE
MSDs. 2 x 2 contingency tables were used to find an association between the hazard
classifications and the morbidity classifications for the 56 right and left hands and the 28
jobs. For the sides, the authors found, the association between hazard classification and
morbidity classification to be statistically significant (OR = 22.0, p < 0.001). The
sensitivity, specificity, positive predictive value, and negative predictive value were 0.86,
0.79, 0.92, and 0.65, respectively. The authors noted similar results for the jobs (OR =
50.0, p = 0.001). The sensitivity, specificity, positive predictive value, and negative
predictive value were 0.91, 0.83, 0.95, and 0.71. These results provide additional
evidence of the external validity and predictive validity of the SI. These results
demonstrate that the SI can effectively identify tasks that do and do not place workers at
an increased risk of DUE MSDs. These results also demonstrate the SI’s external
validity.
Stephens, Vos, Stevens, & Moore, 2006, evaluated the test-retest repeatability of
published data collection and rating methods of the SI by analyzing 61 job video files
twice over a 5-month period. The authors found intraclass correlation coefficients for
task variable ratings and accompanying data ranged from 0.66 to 0.95 for both
individuals and teams. The authors also found SI Score intraclass correlation coefficients
for individuals and teams were 0.56 and 0.82, respectively. Intra-rater reliability for the
hazard classification was 0.81 for individuals and 0.88 for teams. These results suggest a
good test-retest repeatability for the SI.
22
TLV for HAL. TLV for HAL is based on two variables, which are hand activity
and normalized peak hand force. Hand activity level (HAL) is represented as a numerical
score (0-10) and can be either referenced from a table based on frequency of exertion
(efforts per minute) and duty cycle (percent duration of exertion), or from a verbal anchor
scale (Latko, Armstrong, Foulke, Herrin, Rabourn, & Ulin, 1997). Normalized peak
hand force (NPF) can be calculated by using EMG, or estimated using the Borg CR-10
rating, and is expressed on a 0-10 scale. Two limits are defined with TLV for HAL,
using peak force and HAL rating; they are the action limit (AL) and the threshold limit
value (TLV) (American Council of Governmental Industrial Hygensists, 2002). HAL
and peak force are plotted on the TLV for HAL evaluation graph. If the job falls above
the TLV line on the graph, it is said to be hazardous to most workers. If the plotted point
falls below the AL line, the job is said to be “safe” to most workers. Jobs falling between
the TLV and AL line are at moderate risk. The TLV for HAL is only intended to be used
for “mono-task jobs”, where similar motions are performed repeatedly for four or more
hours a day.
A cross-sectional study by Franzblau, Armstrong, Werner, and Ulin (2005) used
908 workers from multiple job sites to analyze prevalence of symptoms and upper
extremity disorders with the TLV for HAL. The authors categorized workers exposures
as above the TLV, above the AL but below the TLV, or below the AL. The authors
found that all measures of CTS (X² = 4.34, p = 0.037) and elbow and forearm tendonitis
(X² = 11.68, p = 0.0006) were significantly associated with TLV category. The authors
found that symptoms in the DUE and wrist, hand, and finger tendonitis did not vary by
TLV category. The authors note that some symptoms and specific disorders occurred
23
below the AL, indicating that even with “acceptable” levels of hand activity, workers
may develop symptoms and/or disorders. These results suggest limited support for the
effectiveness and validity of the TLV for HAL. However, the authors found TLV
categories were positively associated with both elbow/forearm tendinitis and diagnosed
CTS.
Bonfiglioli, Mattioli, Armstrong, Graziosi, Marinelli, Farioli, & Violante (2012)
evaluated the risk of CTS using the TLV for HAL. There were 3,860 participants who
had completed all baseline criteria. The authors found that the TLV classification
predicted both CTS symptoms (IRR between AL and TLV 2.43, 95% CI = 1.77 – 3.33;
IRR above TLV 3.32, 95% CI = 2.34 – 4.72) and CTS confirmed by nerve conduction
studies (IRR between AL and TLV 1.95, 95% CI = 1.21 – 3.16; above TLV 2.70, 95% CI
= 1.48 – 4.91). These results demonstrate support for the effectiveness of the TLV for
HAL when analyzing CTS.
In addition to studies looking solely at the SI or the TLV for HAL, there have
been a few studies that analyze both. A study by Spielholz, Bao, Howard, Silverstein,
Fan, Smith, & Salazar (2008) evaluated both job analysis methods using 567 participants
from 12 companies in the manufacturing and health care industries. The authors
performed inter-rater reliability comparisons on 125 selected cyclic tasks, with one
novice and three experienced raters. HAL hand repetition ratings had a Spearman r value
of 0.65 and a kappa value of 0.44 between raters. Subjective force estimates had a
Spearman r = 0.28 and were not significantly different between raters (p > 0.05). The
rating comparison for the four subjective components of the SI had Spearman r
correlations of 0.37 – 0.62 and kappa values of 0.25 – 0.44. The SI and TLV for HAL
24
agreed on exposure categorization 56% of the time. Logistic regression showed, after
adjustment for age, gender, and BMI, that higher peak hand force estimates (OR = 1.14,
95% CI = 1.02 – 1.27), most common force estimates (OR = 1.14, 95% CI = 1.02 – 1.28),
hand/wrist posture rating (OR = 1.71, 95% CI = 1.15 – 2.56), SI scores ≥ 7 compared
with ≤ 3 (OR = 2.33, 95% CI = 1.20 – 4.53), and SI scores ≥ 7 compared with < 7 (OR =
1.82, 95% CI = 1.04 – 3.18) were associated with distal upper extremity disorders in the
dominant hand. HAL repetition ratings ≥ 4 (OR = 2.81, 95% CI = 1.40 – 5.62) and
hand/wrist posture ratings (OR = 1.59, 95% CI = 1.01 – 2.49) were associated with
disorders in the nondominant hand. Therefore, these results show moderate to good
inter-rater agreement and significant relationships to health outcomes.
Another study by Garg, Kapellusch, Hegmann, Wertsch, Merryweather, Deckow-
Schaefer, Malloy, & the WISTAH Hand Study Research Team (2012) analyzed both the
SI’s and TLV for HAL’s ability to predict risk of developing CTS. The authors used a
cohort of 536 workers from 10 manufacturing facilities. The workers were followed
monthly for six years. The authors found multiple factors that predict the development of
CTS, which include: job physical exposure (measured by TLV for HAL and the SI), age,
BMI, other MSDs, inflammatory arthritis, gardening outside or work and feelings of
depression. In the adjusted models, the TLV for HAL and the SI were both significant
per unit increase in exposure with hazard ratios (HR) increasing up to a maximum of 5.4
(p = 0.05) and 5.3 (p = 0.03), respectively; however, both suggested relatively lower risk
at higher exposures. The results from this study suggest that the TLV for HAL and the SI
are useful ways of estimating exposure to physical exposure risk factors.
25
The results from the existing literature regarding the SI and the TLV for HAL
show that both have been shown to be somewhat reliable and valid. When analyzing the
literature, the TLV for HAL seems to be less proven than the SI. One reason for this may
be that the TLV for HAL has not been studied as thoroughly as the SI. However, both
seem to demonstrate good external validity and high sensitivity and specificity.
26
Methods
Description of Data Obtained for Current Research (Parent Study)
Brief Description of Parent Study. The parent study was a longitudinal study of
1,205 volunteer workers from 21 manufacturing companies located in IL, UT, and WI.
These workers perform various activities including: poultry processing, manufacturing
and assembly, small electric motor manufacturing and assembly, metal automotive
engine parts manufacturing, and plastic and rubber automotive engine parts
manufacturing and assembly. Workers participated in the study for up to 6 years.
Physical exposure and health outcomes data were quantified at baseline, and re-assessed
at regular intervals throughout the study.
Baseline Health Data – Description. Data were collected through a
questionnaire, structured interview, physical examination, and a nerve conduction study
(NCS). A trained occupational therapist administered the questionnaire and the
structured interview. The questionnaire included demographic, individual, and
psychosocial data. The structured interview assessed the presence of symptoms of
numbness/tingling and/or pain in the distal upper extremity (DUE). In addition, the
baseline structured interview included assessment of the history of specific disorders and
treatments (e.g. CTS, CTS release, etc.). Symptoms and history of disorders were
recorded for each hand separately. A comprehensive physical exam was performed on
each participant by the same therapist that conducted the structured interview. During the
physical examination of the neck to hand regions, the therapist conducted palpation,
performed physical maneuvers, and measured height and weight to calculate body mass
27
index (BMI). A second, confirmatory physical exam was administered by an
occupational medicine physician. Regardless of symptoms, all workers underwent a
nerve conduction study (NCS) of each hand at baseline. These were conducted by a
board certified physiatrist who was blinded to the workers’ symptoms and job physical
exposures. The physiatrist classified workers as having a “normal” NCS, or an
“abnormal” NCS, consistent with median mononeuropathy at the wrist (a detailed health
outcomes methodology is available in Garg, Kapellusch, Hegmann, & Merryweather,
2010).
Baseline Job Data – Description. Job physical exposure data were assessed for
each hand separately on a per worker basis by trained ergonomic analysts. Job data were
obtained through interviews with the workers and their supervisors, by observation, by
measurement, and by video analysis. Numerous job physical exposure data were
collected including: (i) estimated hand forces (Borg CR-10 scale, Borg 1982), (ii)
number of exertions per minute, (iii) duration of hand exertions per cycle and length of
work shift, (iv) hand/wrist posture, (v) speed of work, and (vi) duration of exposure per
day. Analyst overall force rating, frequency and duration of exertions, Hand Activity
Level (HAL) rating, hand/wrist postures, and speed of work were measured using a
verbal anchor scale (Latko, 1997; American Council of Governmental Industrial
Hygenists, 2002). See Appendix A for specific job data forms used to collect baseline
job data.
Follow-up Assessment of UED Health – Description. A trained occupational
therapist visited each worker each month to monitor existing symptoms and to determine
if new symptoms developed during the preceding month. This was done through a
28
structured interview at the participant’s work station. If the participant experienced
new/changed symptoms, a new, partial physical exam was completed by the therapist.
Every six months, those workers who had symptoms consistent with CTS received a
follow-up NCS.
Follow-up Job Data. Every quarter a trained ergonomics analyst, visited
workers at their workstations to determine if the worker was performing the same job or
if he/she was assigned to a new or different job. If the worker was determined to have a
job change, the analyst studied the new job in the same manner as they did at baseline.
Methodology of Current Research
Determination of Study Cohort. Subjects for this study were drawn from the
1,205 workers who participated in the above described parent study. The health outcome
of interest was incidence of aggregate DUE MSDs, defined as: (i) CTS, (ii) lateral
epicondylitis, (iii) medial epicondylitis, (iv) hand/wrist tendonitis, (v) de Quervain’s
disease, and/or (vi) trigger finger. Incidence of aggregate disorders were compared to
physical exposure quantified using the SI and TLV for HAL. The unit of analyses was
the individual worker and analyses were performed at the “person-level”. That is, a
health outcome could occur in the left, right, or both arms, and physical exposure was the
greater of left/right. Eligible workers were those who: (i) underwent complete health
baseline measurement, (ii) underwent complete job baseline measurement, (iii) had
quantified job physical exposure quantified (i.e. video analysis), and (iv) received one or
more monthly follow-up measurements. Those workers who met a specific DUE MSD
case definition for an aggregate disorder at baseline, who previously had an aggregate
29
DUE MSD, or who were ineligible to become a case for one or more aggregate DUE
MSDs were excluded. Those persons reporting symptoms due to an acute injury (e.g.
accident) were excluded by censoring them as a non-event one day before reporting the
symptoms.
Computation of ‘Job Metrics’ at the Task and Job Levels. For this study,
workers were considered to be performing one job at a time. Workers could change jobs
throughout the study, triggering a job re-assessment. Each job consisted of one or more
tasks (e.g. machine operator, assembly worker, etc.) and each task consisted of one or
more sub-tasks (e.g. install screws, paint parts; see Figure 1).
Figure 1. Example of how the breakdown of job, tasks, and subtasks occurred for each
worker.
30
Determining Scores for the Strain Index and TLV for HAL. SI assessments
of sub-tasks were performed using frame by frame video analysis based to the SI
methodology (Moore and Garg, 1995). Analysts estimated force rating (Borg CR-10),
number of efforts/min, duty cycle of efforts (% duration of exertion), hand/wrist posture,
and speed of work were recorded. SI sub-task ratings were summarized at the task level
using the following protocol. Overall intensity of exertion was determined using an
algorithm developed by Drs. Garg and Kapellusch (Appendix B). The algorithm rating
of force, as well as the analyst rating of overall force were used to determine amount of
intensity of exertion. Total efforts per minute and total duty cycle were determined by
summing all sub-task measurements. Hand/wrist posture and speed for the task were
defined as those occurring most often during sub-tasks (by percentage of time).
TLV for HAL score was defined as [analyst Peak Force Rating on Borg CR-10
scale ÷ (10 – HAL Rating)]. Analyst Peak Force rating was measured from sub-tasks.
HAL Rating was estimated using the HAL verbal anchor scale and the rating was
provided at the task-level.
For this study, two methods were used to quantify physical exposure at the worker
level for complex jobs (i.e. jobs with two or more tasks): (i) peak exposure, and (ii)
typical exposure. Peak exposure referred to the task with the highest score (TLV for
HAL and SI separately). Typical exposure referred to the job performed for the greatest
percentage of time. In the event that two tasks shared typical exposure (e.g. a tie in
duration), the task with higher physical exposure was chosen.
31
Occupational UED Health Outcomes – Case Definitions. A worker could have
an aggregate disorder either in the left, right or both hands for the person level analyses
used in this research. Those workers meeting the case definition at baseline, or who had
previously been diagnosed as having an aggregate disorder were excluded from eligibility
for becoming an incident case.
Specific case definitions were used to diagnose aggregate disorders. These case
definitions remained the same throughout the study and reflect the case definitions used
in the parent study’s technical report (Garg, et. al, 2010). Specific case definitions are
provided in Table 1. Workers were considered a “case” upon meeting the criteria for one
or more specific case definitions. Workers reporting that they received medical treatment
(e.g. injection, surgery) for one or more aggregate DUE MSDs became an incident case
at the time they received treatment. Workers who developed symptoms of a specific
DUE MSD due to an accident or an acute injury (e.g. fall, laceration) were right censored
(and recorded as a non-case) one day prior to the accident.
32
Disorder Symptoms Maneuver/ Measurement
Exclusions & Right Censor Conditions
CTS 1. Numbness/tingl ing in 2 or more median nerve served digits (1-4) for at least 2 consecutive monthly follow-up interviews plus abnormal NCS
- Abnormal NCS (time difference between +NCS and consecutive N/T follow-ups ≤ 6 months)
- Evidence of systemic neuropathy - Prior diagnosis of CTS by a Physician -History of a carpal tunnel release - Amputation of second or third digit at MCP or PIP in either hand
Lateral Epicondylitis
1. Lateral elbow pain for ≥ 50% days on monthly follow-up interview
- Pain upon palpation of 1 or more of 6 lateral tender points (from monthly follow- up physical exam) - Positive resisted wrist extension (this test is positive if it elicits pain over the lateral epicondyle)
- Prior diagnosis of lateral epicondylitis - Prior elbow surgery of unknown type and/or injection - Prior radial nerve pain
Medial Epicondylitis
1. Medial elbow pain for ≥ 50 percent days on monthly follow-up interview
- Pain upon palpation of 1 or more of 2 medial tender points (from monthly follow- up physical exam) - Positive resisted wrist flexion (this test is positive if it elicits pain over the medial epicondyle)
- Prior diagnosis of medial epicondylitis - Prior elbow surgery of unknown type and/or injection - Prior ulnar neuropathy or cubital tunnel surgery, OR clinical impression of ulnar neuropathy
33
Table 1 (cont.): Specific case definitions for aggregate disorders cont.
Extensor Tendonitis
1. Dorsal wrist pain for ≥ 50% days on monthly follow-up interview 2. 2-6 extensor compartment tenderness (dorsal wrist area)
- Positive resisted wrist extension (this test is positive if it elicits pain over the lateral epicondyle)
- History of wrist arthritis - Prior diagnosis of extensor tendonitis - Prior surgery for extensor tendinitis - Prior wrist surgery or injection of unknown type Right Censored: - Develops wrist arthritis
Flexor Tendonitis
1. Volar wrist pain for ≥ 50% days on monthly follow-up interview 2. No numbness/tingling in digits 1-4 from monthly follow-up interview
- Three locations of digital flexor tendon tenderness from monthly follow-up physical exam - Positive resisted wrist flexion (this test is positive if it elicits pain)
- Prior diagnosis of flexor tendonitis - Prior surgery for flexor or extensor tendinitis - Prior wrist surgery or injection of unknown type - History of wrist arthritis Right Censored: - Develops wrist arthritis
De Quervain’s Disease
1. Radial wrist pain (thumb side) for ≥ 50 % days on monthly follow-up interview
- 1st extensor compartment tenderness (base of thumb) from monthly follow-up physical exam - Positive Finkelstein test (active) from monthly follow-up physical exam
- Prior deQuervain’s diagnosis - Prior deQuervain’s surgery - Hand surgery or injection of unknown origin - History of CMC/Wrist/MCP arthritis Right Censored: - Develops CMC/Wrist/MCP arthritis
Trigger Finger
1. Pain in the finger from monthly follow-up interview and focal tenderness over A-1 pulley (close to the MCP joint) from physical exam
- Demonstrated triggering (a catching of a digital flexor tendon as it glides under the A-1 pulley) from monthly follow- up physical exam OR monthly interview
- History of trigger finger/thumb - Prior finger/hand surgery - MCP/finger osteoarthritis at baseline
34
Statistical Analyses
Statistical Modeling. Typical exposure was used for analysis, since this is the
exposure the worker experiences the most. Time to first event of aggregate disorders was
modeled using proportion hazards (PH) regression (Cox, 1972). Incident cases were
censored from the timeline on the day they met the case definition for one or more
aggregate disorders. Workers, who left the study prior to becoming an incident case,
were censored at the time they left the study (and were non-cases).
Since physical exposure could change throughout the study, the SI and TLV for
HAL were treated as time-varying covariates within the model. Age, gender, and BMI
were treated as time-independent variables using baseline values.
Separate models were created for SI and TLV for HAL using both their
continuous forms and pre-defined categories as reported in Moore, Vos, Stephens, &
Garg (2006) for SI, American Council of Governmental Industrial Hygienists (2002) for
TLV for HAL. Age, gender, and BMI were used as covariates in both models. All
statistical analyses were performed in R-64 version 2.13.1 (R Development Core Team,
2011).
Determining Functional Form of Continuous Variables. The function form of
the SI, TLV for HAL, age, and BMI were determined by fitting the null PH model of
aggregate disorders and plotting the Martingale Residuals of that model against each
variable separately. (Therneau, Grambsch, & Fleming, 1990) Cubic spline smoothing
curves were used to display the functional form of the variables.
35
The variables showing a linear relationship between the variable and incident
cases, were modeled as linear functions. The variables suggesting a non-linear
relationship, were transformed using linear splines with a single knot. The knot of the
linear spline was placed at the nearest quantile of cases to the inflection point on the
function form.
Descriptive Statistics
Enrollment. Workers were recruited during the first 18 months from the
beginning of the study. Out of 673 workers initially enrolled at baseline, 552 workers
(82%) completed job baseline data collection and 667 (99%) completed the health
baseline data collection (Figure 2). A total of 552 (82%) workers completed both job and
health baseline data collections and of these 536 (97%) completed one or more months of
follow-up. Of these workers, 264 (49%) were eligible to participate in the study and 272
(51%) were ineligible to participate. Over the six-year follow-up period, 69 of the 264
workers became incident cases (26%).
Prevalence. Point and lifetime prevalence for each of the six aggregate DUE
MSDs within the cohort of 536 workers were calculated and are provided in Table 2. For
CTS, baseline prevalence was 10% and life prevalence was 20%. Point prevalence for
lateral epicondylitis was 5% and life prevalence was 15%. For medial epicondylitis,
point prevalence was 1% and life prevalence was 4%. Point prevalence for de
Quervain’s disease is 5% (25 cases: 1 male, 24 females) and lifetime prevalence is 5%
(28 cases: 1 male, 27 females). For trigger finger, point prevalence is 12% (63 cases: 12
males, 51 females) and life prevalence is 25% (132 cases: 38 males, 94 females). Point
prevalence for extensor tendinitis was 9% and 1% for flexor tendinitis. Life prevalence
for both flexor and extensor tendinitis was 18% (94 cases: 19 males, 75 females).
37
Table 2: Point and lifetime prevalence for the six aggregate DUE MSDs
Specific Aggregate Disorders
94/18%
132/25%
Number of cases/Percentage of cases out of entire cohort (N = 536)
Occurrence of Aggregate Disorders. Among the 69 workers who became
incident cases, 24 (35%) developed two disorders and three (4%) developed three
disorders. Lateral epicondylitis was the first disorder to occur in 19 (28%) cases,
followed closely by trigger finger, which occurred first in 17 (25%) cases and CTS,
which occurred first in 12 (17%) cases (Table 3). Trigger finger was the second disorder
among 8 of the 24 (33%) workers who developed two or more disorders. All other
second disorders occurred about equally. Among the incident cases, trigger finger and
lateral epicondyltiis most commonly occurred with 25 of 69 (36%) and 23 of 69 (33%) of
workers developing these disorders respectively. CTS occurred in 15 of 69 workers
(22%). The remaining disorders occurred in 10 of 69 (14%) of workers or less each.
38
Disorders
Number of Cases that
Developed as a Second
Total Occurrences
Covariates. Demographics of the total cohort, virgin cohort (incident eligible)
and prevalent cohort (not incident eligible) are provided in Table 4. The mean age of the
total participants was 42.16 (eligible 39.67, ineligible 44.57). The mean BMI for total
participants is 29.09 (eligible 28.27, ineligible 29.90). Of the total cohort, 361 (67.4%)
are female and 175 (32.6%) are male (eligible females = 155 (58.7%), males = 109
(41.3%), ineligible females = 206 (75.7%), males = 66 (24.3%)).
39
Variable Category n Percentage or Mean ± Standard Deviation
(range)
Continuous Continuous Continuous
536 264 272
42.16 ± 11.55 (18.7 – 68.1) 39.67 ± 11.95 (18.7 – 68.1) 44.57 ± 10.64 (19.3 – 68)
BMI at baseline BMI (total) Eligible Ineligible
Continuous Continuous Continuous
536 264 272
29.09 ± 6.81 (16.5 – 58.6) 28.27 ± 6.16 (16.5 – 54.9) 29.9 ± 7.31 (16.6 – 58.6)
Gender (total) Eligible Ineligible
Physical Exposure Variables. Descriptive statistics physical exposure variables
are provided in Table 5. The mean intensity rating for the analyst SI for the entire cohort
is 2.37 ± 0.88 (0.5 - 5) (eligible M = 2.33 ± 0.82 (0.5 – 5), ineligible M = 2.41 ± 0.94 (0.5
– 5)). The mean intensity rating from the algorithm SI for the entire cohort is 2.49 ± 1.18
(0.5 – 10) (eligible M = 2.51 ± 1.1 (0.5 – 7), ineligible M = 2.48 ± 1.25 (0.5 – 10)). The
mean score for the algorithm SI with the entire cohort is 17.3 ± 19.36 (0.75 – 234)
(eligible M = 16.41 ± 14.41 (0.75 – 81), ineligible M = 18.16 ± 23.16 (0.75 – 234)). The
mean SI intensity score for the entire cohort is 15.26 ± 13.83 (0.75 – 108) (eligible M =
14.42 ± 13.21 (0.75 – 108), ineligible M = 16.08 ± 14.38 (0.75 – 108)).
40
The mean analyst TLV rating for the entire cohort is 0.87 ± 0.62 (0.07 – 6)
(eligible M = 0.83 ± 0.59 (0.1 – 6), ineligible M = 0.90 ± 0.64 (0.07 – 4)). The mean
worker TLV rating for the entire cohort is 1.04 ± 0.84 (0 - 7) (eligible M = 0.98 ± 0.74
(0.13 – 5), ineligible M = 1.09 ± 0.93 (0 – 7)). Additionally, the mean efforts per minute
for the entire cohort is 26.19 ± 14.53 (0.8 – 98.3) (eligible M = 26.49 ± 15.2 (1.6 – 98.3),
ineligible M = 25.90 ± 13.87 (0.8 – 69)).
Table 5: Descriptive statistics for physical exposure factors
Variable Category n Percentage or Mean ± Standard Deviation
(range)
Continuous Continuous Continuous
536 264 272
2.37 ± 0.88 (0.5 – 5) 2.33 ± 0.82 (0.5 – 5) 2.41 ± 0.94 (0.5 – 5)
Algorithm SI – Intensity (Typical) Eligible Ineligible
Continuous Continuous Continuous
536 264 272
2.48 ± 1.25 (0.5 – 10)
Continuous Continuous Continuous
536 264 272
17.3 ± 19.36 (0.75 – 234) 16.41 ± 14.41 (0.75 – 81) 18.16 ± 23.16 (0.75 – 234)
SI Intensity – Score (Typical) Eligible Ineligible
Continuous Continuous Continuous
536 264 272
15.26 ± 13.83 (0.75 – 108) 14.42 ± 13.21 (0.75 – 108) 16.08 ± 14.38 (0.75 – 108)
Analyst TLV (Typical) Eligible Ineligible
Continuous Continuous Continuous
536 264 272
0.87 ± 0.62 (0.07 – 6) 0.83 ± 0.59 (0.1 – 6) 0.9 ± 0.64 (0.07 – 4)
41
Variable Category n Percentage or Mean ± Standard Deviation
(range)
Continuous Continuous Continuous
536 264 272
1.09 ± 0.93 (0 – 7)
Continuous Continuous Continuous
536 264 272
26.19 ± 14.53 (0.8 – 98.3) 26.49 ± 15.2 (1.6 – 98.3) 25.9 ± 13.87 (0.8 – 69)
Group Differences. Chi-square and independent samples t-tests were run to
determine significant differences among eligible and ineligible participants. The workers
of the virgin cohort were proportionally more male than female (41.3% male in eligible
vs. 24.3% male in ineligible, X² = 17.66, p ≤ 0.001) and had lower BMI eligible = 28.26,
ineligible = 29.90, p = 0.01). No significant differences were found between eligible and
ineligible workers for age, worker TLV, algorithm SI – intensity, analyst SI – intensity,
algorithm SI score, SI intensity score, and efforts per minute.
Univariate Analyses
Table 6 summarizes the results from univariate analyses of the age, gender, and
BMI covariates. Age (HR = 1.03 (95% CI: 1.01 – 1.05), p = 0.001) and gender (HR =
2.38 (95% CI: 1.36 – 4.17, p = 0.002) were found to be statistically associated with
increased risk of aggregate upper extremity disorder. No increased risk was associated
with BMI (HR = 1.02, 95% CI: 0.99 – 1.06, p = 0.22).
42
Variable (overall p-value)
Age Continuous-linear (per unit increase)
264 (69) 1.03 (1.01 – 1.05) 0.001**
BMI Continuous-linear (per unit increase)
264 (69) 1.02 (0.99 – 1.06) 0.216
Gender Male Female
No statistically significant association with increased risk for aggregate DUE
MSDs was found for TLV for HAL or the SI (p > 0.13) (Table 7). Secondary analyses
found that when transformed using a linear spline, efforts per minute was associated with
increased risk of aggregate disorders (p < 0.03) (Table 7).
Table 7: Univariate hazard ratios for exposure variables
Variable Category/function N (cases) Hazard Ratio2
95% CI p-value
Analyst SI Model
Linear Linear Spline (p=0.43)1
Per unit increase ≤ 9 Per unit increase > 9 Categorical ≤ 6.1 > 6.1
264 (69) 141 (37) 123 (32) 78 (17) 186 (52)
1.01 1.04
0.22 0.46 0.583
Algorithm SI Model
Linear Linear Spline (p=0.96)1 Per unit increase ≤ 9 Per unit increase > 9 Categorical ≤ 6.1 > 6.1
264 (69) 133 (38) 131 (31) 63 (18) 201 (51)
1.00 1.01
0.81 0.83 0.863
0.88
1 Overall p-value for variable transformed as linear spline. 2 Hazard Ratio of 1.0 with no confidence interval indicates reference category for the variable. 3 This p-value is for the second spline term and represents a test for change in slope at the knot. Thus, this p-value
does not correspond to the given confidence interval, which is for the HR beyond the knot point.
43
Variable Category/function N (cases) Hazard Ratio2
95% CI p- value
Analyst TLV for HAL
Linear Linear Spline (p=0.81)1 Per unit increase ≤ 0.75 Per unit increase > 0.75 Categorical < AL AL ≤ score ≤ 0.78 > TLV
264 (69) 199 (41) 65 (28) 61 (17) 99 (24) 104 (28)
1.08 0.75
0.78 – 1.50 0.15 – 3.70 0.27 – 8.35 0.40 – 1.39 0.59 – 2.00
0.64 0.72 0.643
Efforts per Minute
Linear Linear Spline (p=0.02)1 Per unit increase ≤ 37.3 Per unit increase > 37.3
264 (69) 211 (52) 53 (17)
1.01 1.04
0.09 0.005 0.033
SI Force Analyst Rating
Linear Linear Spline (p=0.61)1 Per unit increase ≤ 3 Per unit increase > 3
264 (69) 246 (64) 18 (5)
0.97 0.87
0.85 0.45 0.322
SI Force Algorithm Rating
Linear Linear Spline (p=0.25)1 Per unit increase ≤ 3 Per unit increase > 3
264 (69) 220 (61) 44 (8)
0.83 0.86
0.11 0.35 0.793
1 Overall p-value for variable transformed as linear spline. 2 Hazard Ratio of 1.0 with no confidence interval indicates reference category for the variable. 3 This p-value is for the second spline term and represents a test for change in slope at the knot. Thus, this p- value does not correspond to the given confidence interval, which is for the HR beyond the knot point.
Multivariate Analyses
Cox Proportional Hazards Regression models were analyzed to determine if the
SI was related to increased risk of developing an aggregate disorder after controlling for
confounders. When introduced into the multivariate model of covariates, analyst SI,
treated as a continuous variable, was not statistically associated with increased risk of
aggregate disorder (p = 0.45). Analyst SI with two categories was also not statistically
associated with increased risk (p = 0.50). When the algorithm SI was introduced into the
multivariate model of covariates and treated as a continuous variable, it was determined
to not be statistically associated with increased risk (p = 0.57). When algorithm SI was
44
treated with two categories, it was also not determined to be significantly associated with
increased risk (p = 0.55).
Cox Proportional Hazards Regression models were also analyzed to determine if
TLV for HAL was associated with increased risk of developing an aggregate disorder.
When introduced into the multivariate model of covariates analyst TLV for HAL, treated
as a continuous variable, was not statistically associated with risk of aggregate disorder (p
= 0.98). Analyst TLV for HAL, when treated with three categories, was also found to not
be statistically associated with increased risk (p = 0.36).
When introduced into the multivariate model of covariates efforts per minute,
treated as a continuous variable, was not statistically associated with increased risk of
aggregate disorder (p = 0.40). Efforts per minute, when using a linear spline function (3rd
quartile), approached significance (p = 0.07). See Tables 8-13 for detailed multivariate
model information.
Table 8: Multivariate model for risk of aggregate disorders with analyst SI variable
Variable (overall p-value)1
Category/function N (cases)
264 (69) 1.01 0.99 – 1.03 0.44
Covariates Age (p=0.003)1 Continuous
(per unit increase) 264 (69) 1.03 1.01 – 1.05 0.003
BMI (p=0.28)1 Continuous (per unit increase)
264 (69) 1.02 0.98 – 1.06 0.26
Gender (p=0.004)1 Male Female
109 (16) 155 (53)
0.007
1 Overall significance associated with including each variable in the model using the likelihood ratio test.
45
Table 9: Multivariate model for risk of aggregate disorders with analyst SI variable with 2 categories
Variable (overall p-value)1
Category/function N (cases)
SI ≤ 6.1 SI > 6.1
78 (17) 186 (52)
(per unit increase) 264 (69) 1.03 1.01 – 1.05 0.004
BMI (p=0.28)1 Continuous (per unit increase)
264 (69) 1.02 0.98 – 1.06 0.27
Gender (p=0.006)1 Male Female
109 (16) 155 (53)
0.01
1 Overall significance associated with including each variable in the model using the likelihood ratio test.
Table 10: Multivariate model for risk of aggregate disorders with algorithm SI variable
Variable (overall p-value)1
Category/function N (cases)
Covariates Age (p=0.002)1 Continuous
(per unit increase) 264 (69) 1.03 1.01 – 1.05 0.003
BMI (p=0.27)1 Continuous (per unit increase)
264 (69) 1.02 0.98 – 1.06 0.26
Gender (p=0.003)1 Male Female
109 (16) 155 (53)
0.005
1 Overall significance associated with including each variable in the model using the likelihood ratio test.
46
Table 11: Multivariate model for risk of aggregate disorders with algorithm SI variable with 2 categories
Variable (overall p-value)1
Category/function N (cases)
SI ≤ 6.1 SI > 6.1
63 (18) 201 (51)
BMI (p=0.31)1 Continuous (per unit increase)
264 (69) 1.02 0.98 – 1.06 0.299
Gender (p=0.002)1 Male Female
109 (16) 155 (53)
0.004
1 Overall significance associated with including each variable in the model using the likelihood ratio test.
Table 12: Multivariate model for risk of aggregate disorders with analyst TLV for HAL variable
Variable (overall p-value)1
Category/function N (cases)
Continuous (per unit increase)
Covariates Age (p=0.003)1 Continuous
(per unit increase) 264 (69) 1.03 1.01 – 1.05 0.003
BMI (p=0.29)1 Continuous (per unit increase)
264 (69) 1.02 0.98 – 1.06 0.280
G